---
title: "Getting to Know Your Data: Univariate Description"
author: "Carley Palkon, Casey Zanowic, Suzanna Linn"
output: 
  flexdashboard::flex_dashboard:
    storyboard: true
    social: menu
    source: embed
---

```{r setup, include=FALSE}
library(flexdashboard) #
library(rio)
library(ggthemes)
load("data/SCNominees.RData")
load("data/counties.RData")
df <- import("data/cabinet-turnover.csv")
df$length <- as.numeric(df$length) #length was imported as a character vector, changes to numeric

nominees$Result <- factor(nominees$Result, labels = c("Confirmed", "Delayed", "No Action", "Postponed", "Rejected", "Withdrawn"))
```

### **Levels of Measurement:** Regardless of the analysis you want to conduct with your data, the first question to answer is what is the level of measurement of my variable(s)

![](images/LevelsMeasurement.png)


**** 

-	Nominal: Variables that do not have meaningful quantitative values. Instead the values of the variable denote “labels.”

    +	Gender (Male, Female, Other), 
    + Political Party (Republican, Democrat, Independent, Other)
    + Race (Black, White, Hispanic, Asian, Other)

-	Ordinal: Values of the variable have a natural order or rank, but the differences between those ranks aren’t equal.

    + Levels of Education (High School/College/PhD), 
    + T-Shirt Size (Small/Medium/Large)
  
-	Interval: Values of the variable have a natural order and differences between the values are equal: “Scale.” 

    + Unemployment Rate (percentage of the population), 
    + Vote Share (percentage of the population)
    + GDP per capita


### **Summary Statistics:** How you summarize the main features of a variable depends on the level of measurement






  **Summary Statistics**


****

|                                                                                                                                                                                                                                                                                         |                                                                       |                                                                                                             |
|-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------------------|-------------------------------------------------------------------------------------------------------------|
|                                                                                                                                         Nominal                                                                                                                                         |                                Ordinal                                |                                                  Interval                                                   |
| Central Tendency:
Mode 
 
                                                                                                                                                                                                                                                     | Central Tendency:
Mode or Median
 
                          | Central Tendency
Mean, Median, Mode
*Note: Mode will seldom be useful; median is not influenced by outliers, mean is. *
 
  |
| Dispersion:
Create a Frequency Table
 
                                                                                                                                                                                                                                        | Dispersion:
Create a Frequency Table
 
                      | Dispersion 
Range, Interquartile Range, Standard Deviation
 
                                      |
| Shape of the Distribution:
Unimodal, Bimodal, Multimodal
 
*Note: Be careful when describing the shape of the distribution of a nominal variable because due to it not being coded numerically, the shape can shift depending on how you code the nominal variables*
 
  | Shape of the Distribution:
Unimodal, Bimodal, Multimodal
 
  | Shape of the Distribution
Unimodal, Bimodal, Multimodal, symmetric, Normal
 
                      |
| Visualization:
Bar Plot                                                                                                                                                                                                                                                              | Visualization:
Bar Plot                                            | Visualization: 
Histogram, Box Plot                                                                      |

***

*** 

**Additional Notes: **

- Central Tendency Functions

  +	Mean: calculates the average value: `mean(dataframe$name of variable, na.rm=TRUE)`
  +	Median: calculates the value lying at the midpoint of a frequency distribution: `median(dataframe$name of variable, na.rm=TRUE)`

- Dispersion: 

  +	Range. Calculates smallest ( min[] ) and largest (max[]): `range(dataframe$name of variable, na.rm = TRUE)`
  + Percentiles. The value of the variables that has a certain % of data points below it: Includes interquartile range: `quantile(dataframe$name of variable, na.rm=TRUE, probs=c(insert the percentiles you want))`
  + Standard deviation: tell us average spread of values of a variable around the mean value of a variable.  Empirical rule: 68% of data falls between 1 standard deviation of mean, 95% of data falls between 2 standard deviations of the mean, and 98% of data falls between 3 standard deviations of the mean: `sd(dataframe$name of variable, na.rm = TRUE)`

- Assessing Data Shape 

  +	Symmetric distribution: the left side of the distribution mirrors the right side
  + Left skewed: all the data is concentrated on the RIGHT of the graph: looks like a shoe coming down on the left side. 
  +	Right skewed: all the data is concentrated on the LEFT of the graph: looks like a shoe coming down on the right side. 

***


### **Frequency tables:** Present the number (and proportion) of cases in a data frame for each value of a categorical variable

```{r, include = FALSE, echo = FALSE}
library(dplyr)
```
```{r}
knitr::kable(nominees %>%
  group_by(Result) %>%
  summarise(freq = n()) %>%
  mutate(prop = round(freq / sum(freq),3))
# The overwhelming majority (119 of the 163 or 73%) of nominees to the US Supreme Court are confirmed; 12 or 7.4% were withdrawn or rejected; 10 (6.1%) were not acted upon; 7 (of 4.3%) of nominations were delayed and 3 (just under 2%) were postponed
)
```


To precisely describe the central tendency and dispersion of a categorical variable we typically produce a frequency table, which counts the number of cases for each value of the variable. Often we augment the frequency table by reporting the proportion of cases as well as the counts. The table to the left reports the number and proportion of nominees to the US Supreme Court that were confirmed (119, 0.730), delayed (7, 0.043), not acted upon (10, 0.061), postponed (3, 0.018), rejected (12, 0.074), and withdrawn (12, 0.074) beginning with the nominees of George Washington and ending with those of Donald Trump.

***


```{r, eval = FALSE, echo = TRUE}
library(dplyr)
name of data frame %>%
group_by (name of categorical variable) %>%
summarize (freq = n()) %>%
mutate (prop = freq/sum(freq))

```


### **Bar plots:** Present the information in a frequency/proportion table as a visualization of the central tendency, dispersion, and distribution of a categorical variable 



```{r, message = FALSE, warning = FALSE}
library(plotly)
p <- ggplot(data = nominees, mapping = aes(x = Result)) +
  geom_bar(fill = "red") +
  labs(title = "Nearly All Nominations to the United States Supreme Court Are Confirmed",
       x = NULL,
       y = "Frequency",
       caption = "www.senate.gov")

library(plotly)
q <- ggplot(data = nominees, mapping = aes(x = Result)) +
  geom_bar(aes(y = ..prop.., group = 1), fill = "red") +
  labs(title = "Nearly All Nominations to the United States Supreme Court Are Confirmed",
       x = NULL,
       y = "Proportion",
       caption = "www.senate.gov") 
library(gridExtra)
grid.arrange(p,q, nrow = 2)
```

***

**Creating a bar plot of frequencies**

```{r, message = FALSE, warning = FALSE, eval = FALSE, echo = TRUE}
library(ggplot2)
ggplot(data = name of data frame, mapping = aes(x = factor(nominal variable name))) +
   geom_bar() + 
   labs(title = “Whatever You Want”, x = “Explanatory Variable Name”,
   y = “Frequency”,
   caption = “Source of data”) +
  scale_x_discrete(labels = c("Label Category 1", "Label Category 2", etc))

```

Note: If you have a factor class variable, you don't need to use the `factor()` function when providing the x aesthetic. If it is labeled, you won't need `scale_x_discrete()`.

**Creating a bar plot of proportions**

```{r, message = FALSE, warning = FALSE, eval = FALSE, echo = TRUE}
library(ggplot2)
ggplot(data = name of data frame, mapping = aes(x = factor(nominal variable name))) +
   geom_bar(es(y = ..prop.., group = 1), fill = "Some Color") + 
   labs(title = “Whatever You Want”, x = “Explanatory Variable Name”,
   y = “Frequency”,
   caption = “Source of data”) +
  scale_x_discrete(labels = c("Label Category 1", "Label Category 2", etc))
```

When plotting proportions of a single variable, set the `group` argument to 1!


### **Histograms:** Illustrate the central tendency, dispersion, and distribution of an interval variable

```{r, include = TRUE, echo = TRUE}
ggplot(data = counties, mapping = aes(x = prop_college_grad)) +
  geom_histogram(fill = "firebrick", bins = 60) +
  geom_vline(aes(xintercept = median(prop_college_grad, na.rm = TRUE)), color = "black", linetype = "dashed") +
         annotate("text", x = .3, y = 250, 
                  label = paste0("Median = ", round(median(counties$prop_college_grad, na.rm = TRUE), digits = 3))) +
  labs(title = "In a Typical County the Proportion of College Graduates is \nabout .185 but the Proportion in Some Counties is Unusually High",
       subtitle = "2016",
       y = "Number of Counties",
       x = NULL,
       caption = "Source: US Census")

```

****

```{r, message = FALSE, warning = FALSE, eval = FALSE, echo = TRUE}
library(ggplot2)
ggplot (data = dataframe, mapping = aes (x = interval variable)) +
geom_histogram (bins = how many bins you want) +
geom_vline (aes(xintercept = mean or median (interval variable, na.rm=TRUE)), color = “the color”, linetype = “dotted/dashed”) + 
annotate(“text”, x = 0.00, y= 1 [labels for where you want to place the annotation lines on the “grid”], 
         label=paste0(“Mean or Median” = round(mean or median(dataframe$interval variable, na.rm=TRUE), 3))) +
labs(title=”Whatever You Want”,
x= NULL, 
y=  “Frequency”, 
caption = “source of data”)
```


### **Box Plots:** illustrate the central tendency, dispersion, and distribution of an interval variable

```{r, include = TRUE, echo = FALSE}
ggplot(data = df, mapping = aes(x = "", y = length)) +
  geom_boxplot(fill = "blue4") + 
  labs(title = "Length of Time in Office of Cabinet Officials (Days) 1980-2018",
       y = "Frequency",
       x = NULL,
       caption = "Source: FiveThirtyEight") +
  theme_igray()
```

***

```{r, message = FALSE, warning = FALSE, eval = FALSE, echo = TRUE}
library(ggplot2) 
ggplot(dataframe, aes(y=outcome variable,)) +
geom_boxplot() +
labs(title= ”Whatever You Want”,
        subtitle= ”Additional Info”, 
        x= NULL, 
        y= NULL, 
        caption= ”Source of Data”)

```